|
Activity Number:
|
25
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, July 31, 2016 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Business and Economic Statistics Section
|
|
Abstract #319876
|
|
|
Title:
|
Developing Job Linkages for the Health and Retirement Study
|
|
Author(s):
|
Kristin McCue* and John M. Abowd and Margaret Levenstein and Matthew Shapiro and Ann Rodgers and Nada Wasi and Dhiren Patki
|
|
Companies:
|
U.S. Census Bureau and U.S. Census Bureau/Cornell University and University of Michigan and University of Michigan and University of Michigan and University of Michigan and University of Michigan
|
|
Keywords:
|
probabilistic ;
record ;
linkage ;
retirement
|
|
Abstract:
|
This paper documents probabilistic record linkage to create a crosswalk between jobs reported in the Health and Retirement Study (HRS) and workplaces listed on Census Bureau's Business Register. Linking job records joins variables in separate datasets and permits validating responses and developing missing data imputation models. Identifying a respondent's workplace is valuable for HRS because it allows researchers to incorporate the effects of work environments when studying health and retirement behavior. The linkage uses name and address standardizing techniques tailored to business data that were recently developed in a collaboration between researchers at Census, Cornell, and the University of Michigan. The match does not use the identity of HRS respondents and strictly protects confidentiality of information about respondents' employers. The paper describes creation of a set of human-reviewed pairs and how that set is used to train matching models. It then compares several linking strategies and uses a preliminary sample of matched HRS jobs to illustrate alternative ways to incorporate information on match uncertainty into estimates based on the matched data.
|
Authors who are presenting talks have a * after their name.
